Computational modelling is becoming a pervasive tool both for understanding biostructure-function relationships and most recently to guide the development of bioactive molecules. Biomolecular modelling will clearly play an increasingly important role as it provides a systematic means for sorting through the high levels of complexity inherently present in biomolecular systems. Notwithstanding the exciting and increasing successes with biomolecular modelling, a number of serious questions and issues arise concerning the reliability and full exploitation of the techniques. In the case of predictive reliability, the quality of the modelling results clearly depends in turn on the quality of the input potential functions employed. Secondly, it would be very valuable to generate quantitative measures of structure-function relationships especially for the design of bioactive molecules. The long-term goal of this research is to develop methods to help test and improve potentials used in molecular mechanics and molecular dynamics simulations, and to elucidate biomolecular structure-function relationships through the establishment of structure-potential relations. The specific aim of this project is to develop methods of systematic sensitivity analysis to study the relationships between structural or thermodynamic observables and potentials. A key element of the work is to identify key features of a potential that are truly significant in determining a particular structure or function of a biomolecule. Methods to study structure-structure correlations and structural relaxation pathways will also be developed. The necessary software for the sensitivity analysis studies will be written and tested by applying them to study simple peptides, the protein bovine pancreatic trypsin inhibitor (BPTI), trypsin and trypsinogen, and selected proteins from the Protein Data Bank. While testing the software, useful insights into the principles of protein conformation, and the determinants of biomolecular functions will also be gained.